May 5, 1954
REACTION
OF
LINOLEATE AND LINOLELAIDATE WITH DI-t-BUTYL
m.p. 243-244', separated. The identity was confirmed by mixed melting points. The dark filtrate was vacuum distilled down t o one fourth volume. After five days 1.79 g. of m-dibromotetrol, greyish crystals, m.p. 205-213', was obtained. Oxidation of Tetrols
Cyclohexanetetrol-l,2,3,5 with Periodate.-To
0.0362 g. of the tetrol dissolved in a little water was added 50 ml. of 0.0214 II sodium metaperiodate and water 4 s . 100 ml. The solution was kept at 25.0" and 10-ml. aliquots were withdrawn at intervals for titration. Each aliquot was immediately mixed with 25 ml. of 0.01 N sodium arsenite, 10 ml. of saturated sodium bicarbonate and 1 ml. of 1.2 Jf potassium iodide solution. After 15 minutes standing the excess arsenite was determined iodometrically. The 1,2,3,5-tetrol gave the following results (theoretical consumption 2.0 moles). 0 0.5 1.0 2.0 5.0 22.5 Time, hr. Periodate consumed, moles/mole 0.00 2 . 0 7 2 . 0 7 2 . 1 1 2 . 1 2 2 . 1 8
Cyclohexanetetrol-l,2,4,5with Periodate .-The above procedure was used and tlie following anomalous results obtained (theoretical consumption 2 .0 moles). Ti me, hr . 0 1 . 2 5 6 . 1 8 . 1 2 0 . 5 2 9 . 3 49.6 Periodate consumed, moles/mole 0 . 0 1.50 2.04 2 . 3 2 3 . 4 6 4.14 5 . 3 5
[CONTRIBUTIOS FROM
THE
PEROXIDE
2379
Oxidation with Lead Tetraacetate (Abraham's Carbon Dioxide Method) .-In a control experiment L-arabinose when oxidized with lead tetraacetate using Abraham's procedure13 gave only 7 5 8 7 % of the theoretical carbon dioxide (reported13 99%). The procedure was modified by using a brom thymol bluebrom cresol purple mixed indicator,21 and a sintered glass bubbler in the carbon dioxide absorption tube. This made the procedure more convenient but did not improve the above result. Several runs on the 1,2,3,5-tetrol gave 75-76% of the theoretical one mole of carbon dioxide. The 1,2,4,5-tetrol which one might expect to give no carbon dioxide, gave 2.52 moles in four hours, when the experiment was discontinued.
Acknowledgment.-1I-e wish to thank the National Research Council, the Research Council of Ontario, and the Advisory Committee on Scientific Research for their generous support of this work. We are indebted to the Plantation Division, U. S. Rubber Co., for our supply of quebrachitol, and to Dr. 0. Wintersteiner of the Squibb Institute for Medical Research for samples of inositol dibromohydrins prepared in his laboratory. (21) I. M. Kolthoff and V. A. Stenger, "Volumetric Analysis," 2nd Edit., Vol. 11, Interscience Publishers, Inc., New York, N. Y., 1947, p. 5 8 .
TORONTO, CAXADA
CHEMICAL LABORATORIES, GENERALMILLS,Isc.]
The Reaction of Methyl Linoleate and Methyl Linolelaidate with Di-t-butyl Peroxidel>z BY S. -4. HARRISCN A N D D. H. ~ T H E E L E R RECEIVEDDECEMBER 4, 1953 The reaction of methyl linoleate and methyl linolelaidate with di-t-butyl peroxide a t 125' has been studied. I t was found that the products were largely the dehydro dimers of the fatty esters. The dimer from each fatty ester was found t o be a mixture of isomers differing in the number of conjugated double bonds and in the cis, trans configuration of the double bonds.
Recent reports have shown that when methyl linoleate and di-t-butyl peroxide are heated together a t from 125-135' a dehydro methyl linoleate dimer is formed314accompanied by a minor amount of higher molecular weight material. The dimer has four double bonds of which somewhat less than one-half are conjugated. This dehydro dimer is formed in the stoichiometric proportion of one mole of dimer for each mole of peroxide decomposed. I t is considered to result from the loss of a hydrogen atom a t carbon no. 11 of methyl linoleate, followed by combination of two of the three limiting resonance hybrid radicals. The results of a more detailed study of the reaction of di-t-butyl peroxide and the methyl esters of linoleic and linolelaidic acids are reported in this paper. Decomposition of Di-t-butyl Peroxide.-Examination of the volatile products formed when dit-butyl peroxide was decomposed in the methyl esters showed that t-butyl alcohol was nearly the sole decomposition product of the peroxide. Ad(1) Paper No. 143, Journal Series, Research Laboratories, General Mills, Inc. ( 2 ) Presented in part a t t h e 123rd Meeting of the American Chemical Society in 1.0s Angeles, California. (3) S. A. Harrison and D. H . Wheeler, Minnesota Chemist, 4 , No. 5 , 17 (1952). (4) A. L. Clinginan and U. A. Sultun, J. A w . Oil Cheinisfs' Soc., 30, 33 (1053).
vantage was taken of this fact to follow the decomposition of the peroxide. The alcohol was determined quantitatively and used as a direct measure of peroxide decomposed. I n Fig. 1 the log of the per cent. di-t-butyl peroxide is plotted against reaction time for the decomposition of peroxide in methyl linoleate and methyl linolelaidate. It is evident that the decomposition in the two esters is first order and in agreement with the order and rates (in other media) reported by Raley, Rust and Vaugham6 Dimer Formation.-When low concentrations of di-t-butyl peroxide are used very little ester polymer of higher molecular weight thrtn dimer is formed. For example, when 12 mole per cent. of di-t-butyl peroxide is decomposed in either methyl linoleate or methyl linolelaidate about 95% of the polymeric product is dimer. The amount of dimer was determined by molecular distillation of the reaction mixture concurrently with the determination of the peroxide decomposed. The results which are plotted in Fig. 2 show that very nearly one mole of dimer is formed for each mole of di-tbutyl peroxide decomposed. Structure of Dimers.-As was pointed out by Clingman and Sutton4 the t-butoxy radicals remove hydrogen atoms from carbon atom no. 11 in ( 5 ) J. H Raley, F F.Rust and m' 1s Vaughdn, THISJ O U R N A L , 70, 1336 (1048)
S.
“3SO
x w .‘?
-
.3
1’01. 76
ing the molecular distillation of methyl linoleate polymer mixtures gave evidence that the dimer is in fact a mixture of isomers which vary in the amount of conjugated unsaturation, as shown by the ultraviolet extinction coefficient, K , a t 235 r n p . In order to obtain additional information about the nature of the dimers a large batch (558 g.) of the polymer obtained from safflower methyl esters was distilled in a cyclic niolecular still. About thirty fractions were taken. The physical constants on representative samples were used to plot the curves shown in Fig. 3. .$bout 10% of monomer remained in the polymer mixtiire which had been previously stripped in a falling film still.
-
n .4
IA
*I.HARRISON AND I). H. L~IIEELER
0
260
.I -.
250
I
I
I
I
I
I
I
5
IO
15
20
25
30
35
-
1
U.V. ABSORPTION
HOURS.
1pig. 1.--L)cconiposition of tii-t-butyl peroxide iii inethyl linoleate (0)and Inethyl 1iiiolelaid:ite ( 0 )a t l%fio.
K
e
! 1.4900 1.4860
EI 3 5 -
1.4820
K
2
30-
U
1.4740
$
25-
g
20-
a
Ib
-I
0
3 -1
2
15
-
IO
-
40 o: do 710 PERCENT D I S T I L L E D .
5
IO
15
20
25
35
30
40
MILLIMOLES OF PEROXIDE DECOMPOSED. l-ig 2 -Dimer formation wrsus peroxide decomposition at 1%’ (methyl linoleate 0, methyl linolelaidate 0 ) .
the linoleate ester to give a radical (A). This has thc limitinq hybrid forins (€3) and (C) I:
( 1 1 - C I I -CII
A I)
)r-C P O ‘ 0 CH1
There are six isomeric dimers which are possible t h r o u q h tlir comhinrition of the xppropriate radi c J s ’1hese would differ in the number of coiijugated tlouble bonds and the positioii of the double
bonds.
100
Fractioiis takcn a t diffcrcnt intervals dur-
:tiid
The dimer represents the next GOY0 distilled. There is considerable variation in the properties of the different fractions. As the distillation progresses, the index of refraction and the extinction coefficient at 233 rnp increase until all of the dimer is distilled. the trimer begins to come over, the extinction coefficient drops rapidly whereas the index of rcfraction continues t o rise. The increase in the refractive index through the dimer plateau can be attributed to the increase in the proportion of conjugated material in the dimer fractions as shown by the increase in the extinction coefficient. In the trimer fraction the increase in refractive index is probably partly due to cyclization during clistillation since it was fouiid that cyclization docs occur a t these temperatures. Additional evideiicc of cyclization is t h a t the trimer fractions showed a tlecrease in unsaturation. Clingman and Suttori found no such decrease in their undistilled fraction which should have been largely trimer. The larger proportion of trimer and higher molecular weight material in this preparation results from using about tiuuble the amount (2-1 mole per cent.) of dit-butyl peroxide that had been used in the other preparations. Table I shows analyses of selected fractions from the distillation (rf. Fig. 3 ) . The dimer fractions all show about go(;/;, of the theoretical value for two double bonds pcr linoleate unit or four per dimer.
7
R
do i o
Vig. :;.--3Iolecular distillation of methyl linoleate tliincr trimer inade from safflower methyl Pstcri.
0
w
i o n:
hlay 5 , 1854
REA4CTIONO F LINOLE.\TEAND LINOLEI,AID.ITE WITH DI-t-BUTYL P E R O X I D E
23SI
either carbon 8 or 14, to give free radicals which TABLE I CHARACTERISTICS OF SELECTED FRACTIONS OF DISTILLED could resonate with either 8,12 or 9,13 diene radicals : DIMERA N D TRIMER Double 13 12 I t IO 9 8 bonds0 Double Fraction
by Woburn iodine value
bondsa by hydrogenation
-CH=CH-CH2-CH=CH-CH-
Mo1.b wt.
(k)C 238 mp
1.86 554 36 A 1.87 531 44.5 1.85 B 1.83 553 52 1.76 C 1.80 781 35.7 D .. 1.48 a Number of double bond? per methyl linoleate polvmer. * Ebullioscopic method in benzene. extinction coefficient, 1 g./lOOO cc., 1 cm.
f
+
%BOD
1.4788 1.4791 1.4818 1.4849 unit in Spccific
The measured molecular weights, though somewhat lower than theory, characterize fractions A, B and C as dimers and fraction D as trimer. T h a t the linoleate dimers are mixtures of isomers having various proportions of conjugation is evident from the values of k and refractive index shown in Table I. This is in agreement with what would be predicted if the three different resonance hybrids could combine. Fractions A, B and C were analyzed by infrared to determine cis, trans configurations of the double bonds.6 Fraction A had the most cis, trans conjugated diene (doublet a t 918 and 982 ern.-') while fraction C was richest in trans, trans conjugated diene (strong band a t 988 cm.-’, but only a weak band a t 948 crn.-’). Isolated trans double bonds (968 cm.-l) were present in all three fractions, decreasing from A to C. The presence of an appreciable proportion of isolated trans double bonds shows t h a t during the reaction some cis double bonds have isomerized without conjugating since the original linoleate was from a natural source and contained no isolated trans double bonds. The formation of a trans double bond in the conjugated diene is similar to t h a t observed in the autoxidation of linoleate to hydroperoxide wia the same initial free radical, but the conversion to conjugated diene is less extensive in the present case. The methyl linolelaidate dimer is similar to the methyl linoleate dimer. Infrared analysis shows a large proportion of trans, trans conjugated diene, and isolated trans, with minor proportions of conjugated cis, trans diene. There has evidently been a small amount of isomerization of trans double bonds to cis double bonds. Alkali Isomerization of Dimer.-It has been confirmed, as noted by Clingman and S ~ t t o nt,h~a t the linoleate dimers do not appreciably conjugate with alkali a t 180°, as does normal linoleate. Furthermore, extension of the reaction time did not produce much more conjugation (Table 11). TABLE I1 ALKALIISOMERIZATION OF DIMER Time, min. 0 25 120 360 (235 mp) 35.7 39.5 42.7 42.0
This failure to conjugate extensively with alkali may be due to substitution on the 11-carbon, as proposed by Clingman and S ~ t t o n . Another ~ possibility is that a hydrogen atom is abstracted from ( 6 ) J. E. Jackson, R. F. Paschke, W. Tolberg, H. M. Boyd and D. H. Wheeler. J . A m . Oil Chemists’ Soc., 29, 229 (lU52).
D
13 12 11 10 9 -CH=CH-CH2-CH-CH=CHE
8
Dimers formed from this 1,S-diene type radical (E) would presumably not conjugate with alkali a t 180’. Attack a t the singly activated positions such as Cg and C14 as well as the doubly activated C1l position is apparently possible under the conditions used here. For example, a 1-1 mixture of oleate and linoleate was similarly treated. The resulting dimer contained considerable oleate, as indicated by hydrogen absorption equivalent to 1.;i3 double bonds per monomer unit, compared to 1.94 in the dimer from pure linoleate. Both of these explanations may be correct. The extent to which the second mechanism occurs could be diagnosed by oxidative cleavage of the dimer, or preferably its maleic anhydride reaction product. Reaction with maleic anhydride would presumably occur by Diels-Alder reaction only with the conjugated (9,11 or 10,12) double bonds in the dimers, and would eliminate the formation of azelaic and heptanoic acids from these conjugated portions of the dimer on oxidative cleavage. Any azelaic or caproic acid would then presumably arise from non-conjugated linoleate portions which are attached a t the Cn position, while valeric and suberic acids would be evidence of attachment a t the Cs and C14 positions, as suggested above. Work along this line is planned. Relation of Thermal Dimers to Dehydro Dimers. -Thermal dimers of methyl linoleate differ from the dehydro dimers in two important respects : 1. The thermal dimer has only two double bonds per mole, while the dehydro dimer has four. 2. The thermal dimer has no conjugated double bonds while the dehydro dimer hm about one pair per mole. Farmer proposed a free radical chain mechanism’ for thermal polymerization. This theory involved the formation of a free radical a t the 11-carbon. This radical formed a conjugated diene radical by resonance. The conjugated diene radical added to a double bond of a normal linoleate to form a substituted cyclohexene dimer radical by a “cyclizing addition.” The dimer stabilized itself by removal of a hydrogen atom from a methyl linoleate tn establish the chain reaction. According to this theory, a source of free radicals should catalytically accelerate the formation of this type of dimer. Actually, however, the t-butoxy free radical produced a dehydro dimer which was grossly different from the thermal dimer, as discussed above. Further, the dehydro dimer is formed in stoichiometric proportions, to peroxide, with no evidence of a chain reaction. These facts discredit the Farmer Theory. (7) E, H. Farmer, J . Oil
e Coloirv ChPmisI?’ A.T?,JC.. 31, 393 (1818).
It might be argued that the dehydro dimer is a n intermediate in the formation of thermal dimer, and that the usual thertiinl dimer would result from intramolecular cyclization of dehydro dimer. Heating of the dehydro dimer a t heat-bodying temperatures caused loss of conjugation and unsaturation. However, the dimer thus formed was not identical with the thermal dimer of linoleate. I t had a higher refractive index, probably due to two rings in its structure, coinpared to one ring in the thermal dimer. Experimental Methyl Linoleate.-This ester was prepared by R . F. Paschke of this Laboratory by treatment of the methyl esters of safflower fatty acids with urea. The product had the following constants: ?ZmD 1.4577, iodine value (rapid ii'ijs) 170.3, extinction k (g./l., 1 cm.) 87.0 max. 233 m r after 25 minutes alkali isomerization with KOH-glycol a t 18O0.* These values are close to those reported for the pure methyl cis-Q-, cis-12-lin0leate.~ Infrared analysis indicated the absence of isolated trans double bonds. Methyl linolelaidate was the same as that previously described.6 Di-t-butyl peroxide obtained from the Shell Chemical Corporation had a refractive index n% 1.3890. Reaction of Methyl Linoleate with Di-t-butyl Peroxide.Two hundred grams of methyl linoleate (0.68 mole) and 12 g . of di-t-butyl peroxide (0.082 mole) were placed in a 250cc. flask equipped to collect distillate. I t was heated for 18 hours in an oil-bath held a t 130 =k 0.5'. The system was kept under a slight positive nitrogen pressure (5-7 mm.); After removing all low boiling products by heating t o 130 a t 0.3 m m . the product was distilled in a CMS-5 Centrifugal Cyclic Batch Molecular Still manufactured by Distillation Products Industries. Another batch of 180 g. was similarly treated. The residual trimer from the first preparation was added as a "booster" to that of the second distillation. Ilimer fractions distilling a t rotor temperatures of 160-19,5', .'3--5 u , showed increasing conjugation from k = 35 t o k = 49,c5for the last distilled dimer fraction. This fraction had a saponification equivalent of 295 (theory 293.,5) and a n iodine value (hydrogenation) 168.7 (theory 173.0). Anal. Calcd. for C.l&6604: C , 77.76; H , 11.31. Found: C, 77.33; H, 11.08. The combined low boiling products which amounted to about 19.5 g. (0.263 mole as butanol) were distilled in a Podbielniak concentric tube still. Yirtually the entire product distilled from 81 .0 t o 82.8" a t 250 mm. No acetone was obtained. The phenylurethan of the alcohol melted a t 136.5 to 137.5' (cor.) and showed no depression with another prepared from pure t-butyl alcohol. A sample of the stripped monomer was fractionated in a Podbielniak hlicro High Temperature Analyzer Model 3300. The bulk of the material distilled as unchanged linoleate, with essentially theoretical iodine value and with a k value (after isomerization) only 2-3% low (83-85 as compared to 87 of the original). There was no conjugation before isomerization (ultraviolet and infrared) and trans double bonds were absent (infrared). Reaction of Methyl Linolelaidate with Di-t-butyl Peroxide. -Twenty grams of methyl linolelaidate (0.068 mole) and 1.2 g. of di-t-butyl peroxide (0.0082 mole) were placed in a one ounce small-mouth screw-cap bottle which n-as flushed with nitrogen and then sealed with a metal screw cap equipped with a n oil resistant synthetic rubber gasket. The bottle was heated for 48 hours in an oil-bath held at (8) J. H . Mitchell, Jr., H. R. Krayhill and F. P . Zscheile, I n d . E n g . Chem., Anal. Ed., 15, 1 (1943).
126 40.5'. The mixture was stripped of low boiling reaction products (t-butyl alcohol) by heating utidcr vacuum. Most of the unchanged methyl linolelaidatc was distillctl out of the product by heating to 200' (bath tetnperaturc) a t a pressure of 0.2 mm. in a 100-cc. alembic flask. Part of the stripped dimer was then quantitatively distilled in :I micro molecular still.'O At a pressure of 2 M , the dimer distilled largely in the range 230-260' (block temperature). The distillate amounted to 18.7% of the original reaction mixture. The residue (presumably trimer) amounted to 0.93%. The dimer had the following constants: n3% 1.4813, iodine value (hydrogenation) 169.5 (theory 173.0 extinction coefficient, k = 47.8 max. 235 mp. Anal. Calcd. for C Z ~ H S O C, O ~77.76; : H , 11.31. Found: C, 77.81; H, 11.48. Preparation of Dimer from Safflower Methyl Esters.-A large batch of dimer (1740 g.) was made from safflowcr methyl esters (80% linoleate). For this preparation, double the proportion of di-t-butyl peroxide was used. Monomeric linoleate was removed in a falling film vacuum stilt. A sample (558 9.) of the residual esters was used to carrl- out a large scale molecular distillation in the CMS-5 Centrifugal Still (see Fig. 3 ) . Rate Studies.-The disappearance of the di-t-butyl peroxide was followed by determining the amount of t-butyl alcohol formed, a t 126 i 0.5". The t-butyl alcohol in the mixture was determined from active hydrogen measurements using lithium aluminum hydride. An apparatus similar to that described by Zaugg and Horromg was used. From :I series of known mixtures of methyl linoleate, di-t-butyl peroxide and t-butyl alcohol it was found that it vas necess a r r to subtract 2.7 cc. from the gas volume value for exch 0.1 g . of di-t-butyl peroxide in the mixture. The per cent. of polymer in each sample was determined by molecular distillation of 0.5 to 1 g. of the mixture in a micromolecular still.'O SDectral Ana1vsis.-The absorption maximum of 233-235 tn,u -was determined in a Beckman photoelectric quartz spectrophotometer model D U . Ethyl alcohol (95y0)\ Y , I < used as the solvent. Infrared analyses were run in a Beckman IR-2 spectrophotometer. The samples were dissolved in carbon disulfide (10% by volume). The region 900 t o 1,000 mi.-' was observed6 for absorption characteristic of conjugated t r i i n ~ , trirns, and cis, trans as well as isolated trans double bond,